Hierarchical feedforward models have provided a foundation for most theories of visual processing over the past 40 years. However, in the cerebral cortex there is a dense network of feedback (FB) connections sending topdown information from higher to lower processing centers. The anatomy and function of FB circuits is still poorly understood. This is despite the fact that FB has been implicated in many important visual functions such as contextual modulation, attention and learning. The goal of this application is to understand the anatomical and functional organization of FB connections from the secondary (V2) to the primary visual cortex (V1), and the impact of V2 FB on V1 cell responses. We will test the following hypotheses: 1. Anatomically, V2 FB consists of parallel channels, related to the stripe compartments of V2, with each channel targeting specific V1 layers, V1 compartments and the same V1 cells that provide input to the specific V2 stripe (Aim 1). The different channels also show unique specificities with respect to the functional maps of visual stimulus properties in V1 (Aim 2). 2. Functionally, V2 FB contributes to surround modulation (SM) caused by visual stimulation of the surround regions near the V1 cell?s receptive field (RF) (Aim 3). SM is the ability of stimuli in the RF surround to modulate the responses of V1 cells to stimuli inside the RF, a property thought to serve efficient coding of natural images and segmentation of object boundaries (Nurminen & Angelucci, 2014) Ref43. Our understanding of FB anatomy and function has been hampered by the technical limitations of previous methods used to label FB axons and manipulate their activity. Conventional tracers are either poorly sensitive or label axons bidirectionally, creating a confound in the interpretation of the axon label. As a result, there are conflicting reports regarding the anatomical and functional specificity of FB connections. Inactivation of FB systems has been performed using methods that lack spatiotemporal precision and cell type selectivity, affecting cells in an entire cortical area. These approaches could not rule out that the effects of FB manipulation were mediated by other indirect pathways. To overcome these limitations, we will investigate the anatomical and functional organization of FB connections from V2 to V1, using novel viral-mediated expression of fluorescent proteins to label FB axons and their synaptic terminals unambiguously. In Aim 1 these labeling methods will be combined with staining of V1 layers and compartment (CO maps), and in Aim 2 with optical imaging of V1 functional maps. In Aim 3, manipulation of FB neuron activity will be performed using optogenetic activation of opsin-expressing FB axons, while determining the effects of these manipulations on spontaneous and visually-evoked spike activity in V1 cells, recorded using linear arrays. These data will providean anatomical and mechanistic foundation for modeling and hypothesis-driven studies of FB function and dysfunction, and lead to a better understanding of the computations performed by this canonical cortical circuit, i.e. topdown FB.